Once More unto the Breach: A Reply to Timothy Sandefur’s Response

It appears that Timothy Sandefur disagrees with me. There is certainly a lot of evidence for this in his response to my first posting. However, it is not clear to me how much Mr. Sandefur is disagreeing with me, how much he is disagreeing with Hayek, and how much the disagreement with either of us derives from a semantic confusion. Perhaps some distinctions would help clarify matters.

To begin with, I am willing to classify myself as a Hayekian in the sense that I believe Hayek’s scholarly work to contain several brilliant and incredibly fruitful insights that both advance our understanding of how the world works and supply useful premises in the argument for a free society. I believe his description of spontaneous order to be among these insights. I am neither a Hayek scholar — that would be Bruce Caldwell’s role in this exchange — nor a Hayek disciple — one who believes that Hayek discovered the unvarnished truth and everything he said needs to be defended.

I myself have criticized Hayek’s jurisprudential thinking and understanding of the English customary/common process.[1] I further recognize that Hayek was a rather bad moral philosopher. In his original essay, Mr. Sandefur points out that Hayek asserts that moral principles are a product of an evolutionary selection process while simultaneously and inconsistently employing externally grounded moral principles to “correct” the output of the evolutionary process and criticize rational constructivism. This is a frequently made objection to Hayek’s philosophical thought with which I agreed in my original posting. In addition, I pointed out that Hayek’s commitment to viewing moral principles as a product of evolution leads him to call upon purely formal values such as generality and consistency to do substantive normative work, which they are incapable of doing. Thus, if there is genuine, substantive disagreement between Mr. Sandefur and myself, it concerns the question of whether spontaneous order is a viable concept and whether there is a principled basis for distinguishing spontaneous orders from constructed orders.

The next distinction I’d like to draw is that between orders and actions. The adjectives “spontaneous” and “constructed” are designed to describe two different types of “orders.” An order is a systematic arrangement of items of a particular type or with some common feature or property. A spontaneous order is such an arrangement that arises without any guiding intelligence. A constructed order is an arrangement that is intentionally produced a guiding intelligence. The symmetrical crystalline structure of a snowflake is an example of a spontaneous order. A set of standing dominoes arranged to fall in a snowflake pattern when the first one is knocked over is an example of a constructed order.

These are facile examples of the two types of orders. Why? Because the type of items being ordered — ice crystals and dominoes — are inanimate objects with no wills of their own. Such items are not capable of independent action.

Things become more complex when the items being ordered are human actions. Human beings have free will and can act intentionally. The actions of individual human beings are almost always a product of a guiding intelligence — one’s own. Hence, individual actions are almost always “constructed.” When the items being ordered are human actions, what is being created is an order of intentional (i.e., constructed or planned) actions. Thus, a spontaneous order of human actions is an ordering of individual plans that arises without a guiding intelligence to coordinate them, and a constructed order of human actions is an ordering of individual plans that is itself a product of a conscious plan.

The development of language and the selection of media of exchange are often offered as examples of spontaneous order. In the case of language, individuals intentionally assign meanings to sounds in their dealings with each other, some of which, through repeated interaction, gain general acceptance among a particular community of people, and become a spoken language such as English. This occurs even though there is no overarching guiding intelligence making authoritative assignments of meaning to particular sounds. In the case of media of exchange, individuals select different items as stores of value, some of which, through repeated interpersonal transactions gain general acceptance among a group of people, and become the accepted medium of exchange such as gold generally or hogsheads of tobacco in early Virginia. Of course, both language and media of exchange may be consciously constructed as well, as examples such as Esperanto, Klingonese, the language created for the new movie Avatar, and various forms of government-created fiat money demonstrate.

My working definitions of the terms we are considering, then, are as follows. A spontaneous order of human actions is an ordering of intentionally undertaken or planned individual actions that occurs without the conscious coordination of any guiding intelligence. A constructed order of human actions is an ordering of intentionally undertaken or planned individual actions that is consciously created by a guiding intelligence.

With these distinctions in mind, let’s me see if I can identify the points of disagreement between Mr. Sandefur and myself.


Mr. Sandefur begins his response by stating:

I want to thank Professor Hasnas for so eloquently describing my crucial point: Hayek’s critique of economic or political planning falls short because he “den[ies] any independent grounding for ethics outside of the process of social evolution and… attempt[s] to make purely formal and essentially vacuous values such as generality and consistency do substantive normative work.” Precisely. Hayek’s observations about spontaneous order are useful and important, particularly to counteract some people’s hasty assumption that economic or legal design requires a designer. But that observation must be, so to speak, part of this complete breakfast. Spontaneity itself is not enough; Hayek’s observations must be supplemented by a philosophically rigorous account of justice and human values.

These represent points of agreement between us. However, Mr. Sandefur continues by stating:

But, of course, once we appeal to such values, and use them as “an Archimedean fixed point on which to stand to criticize any of the rules of a spontaneously evolving legal order as unjust and to call for the reform of those rules,” what we are doing is rational constructivism.

I do not believe that this statement follows from what precedes it. It seems to me that in making this assertion, Mr. Sandefur is conflating constructed orders with constructed actions.

We have already agreed that moral principles are not necessarily a product of evolutionary forces, and may be derived from externally grounded, reasoned philosophical positions. Hence, to the extent that any of us perceive that the results of a spontaneous order within which we reside diverge from our conception of justice, we have grounds to criticize them and call for reform. There is a sense in which one can call this “rational constructivism” since my individual position is a product of my own conscious rational design. But this is irrelevant to the question of whether the order within which I make my appeal to justice is spontaneous or constructed. As noted above, when we are ordering human actions, we are always ordering the conscious output of individual human intelligences. The order, as opposed to the actions of the individuals within it, is a spontaneous order just in the case there is no conscious human agency endowed with the authority to make a collective choice. The order is a constructed one if there is. In a spontaneous order, my rationally constructed appeal is successful to the extent it is accepted by other individuals in the order who act upon it. In a constructed order, my appeal is successful to the extent that it influences the agency of collective choice to decide in my favor.

Individuals always function on the basis of their “constructed” beliefs (and sometimes even on the basis of their “rationally constructed” beliefs), but recognizing this fact does nothing to “dilut[e] the distinctions between spontaneous and constructed orders in such a way that it depends entirely on the observer’s point of view; …” This will be a problem only if we hold rigorously to the proposition that ethics is necessarily a product of the spontaneous order process itself — a position that both Mr. Sandefur and myself reject.


I will accept some of the blame for Mr. Sandefur’s second disagreement with me. In my original essay, I carelessly employed the word “final” in describing the essential characteristic of constructed orders as having “a designated final decision maker.” This misleadingly suggests that there is some point in time at which questions are settled and no longer open for review and reform. As Mr. Sandefur points out, constructed orders can change over time as well as spontaneous orders. A more felicitous way of expressing my point would have been to identify the essential characteristic of constructed orders as “having a human agency authorized to make collective choices for the entire order.”

An implication of Mr. Sandefur’s observation is that it can make sense to describe constructed orders as “evolving.” The law of the United States certainly changes in response to the relative success of the various political interests group and rent-seekers who operate within the political sphere. Hence, it is fair to say that the distinguishing feature of spontaneous and constructed orders is not the presence of evolution in one and its absence in another. I did not mean to suggest that it was, but may have inadvertently done so by employing the term “final.”

The presence of some form of evolution in both spontaneous and constructed orders does not imply that there is no principled distinction between them. It may be that the same type of game is being played in each case — there is competition in both spontaneous and constructed orders — but they are being played in different ballparks. Competition in spontaneous orders occurs in the ballpark of individual choice; competition in constructed orders occurs in the ballpark of collective choice.

It is important to appreciate that there can be many different social orders that function at different levels. The question of whether an order is a spontaneous or constructed order is always relative to the particular order under consideration, as is the judgment as to which is more desirable. For example, consider a firm competing in a market within the United States on the planet Earth. If one focuses on the firm, the items being ordered are the employees working for the firm. This is a constructed order because there are identifiable persons authorized to make collective choices that bind the entire firm.[2] If one focuses on the market, the items being ordered are the individuals and firms competing for business. This is a spontaneous order because the arrangement of the order is determined by the individual choices of consumers and no one is authorized to determine who wins and who loses for the entire order.[3] If one focuses on the government of the United States, the items that are being ordered are the citizens of the nation. This is a constructed order because the ordering is done by collective decision-making processes.[4] If one focuses on the planet Earth, the items that are being ordered are the nation-states it contains. This is a spontaneous ordering because there is no agency invested with the power to make collective decisions for the world.

By zooming in and zooming out, Mr. Sandefur claims to show that spontaneous orders can be transformed into constructed orders and vice versa, and hence that there is no principled way of distinguishing them. I do not think he is doing this. I think he is simply transcending levels and looking at different orders. The firm is a constructed order if what is being ordered is its internal components. Zoom out one level and the firm’s rationally constructed actions are simply data points in the spontaneous order of the market. Zoom out another level and you are in a semi-constructed order where the items being arranged are the actions of the citizens of a nation. Zoom out once more and the citizens disappear and the items being arranged are nations-states. Whether one sees a spontaneous or constructed order changes, but that is because one is looking at different orders. In each instance, however, there is no difficulty telling whether the order being observed is a constructed or spontaneous one. This is why Bruce Caldwell can confidently assert that he knows a spontaneous order when he sees one.

A final note on this point. Mr. Sandefur and I are in agreement that spontaneous orders are not desirable merely because they are spontaneous. My argument for a spontaneous legal order — one in which the items being ordered are the rules of interpersonal conduct — is explicitly based on a comparative assessment of the incentives at work in spontaneous and constructed legal orders. Because the incentives in a spontaneous legal order tend to favor peaceful interaction and justice in the long term more than those in constructed legal orders, which grease the path to exploitation, I argue for the former. This argument is not generalizable, however. If the items being ordered are the world’s nation-states, I have reason to doubt that incentives at work in that system would favor peaceful interaction more than those in a constructed world order. On this level, the question whether a spontaneous or constructed order is preferable is an open one.


This brings me to Mr. Sandefur’s and my final area of disagreement. He claims to strongly disagree with my assertion that what Hayek called the rules of just conduct “evolved through customary/common law processes while almost all the rules designed to exploit or oppress others originated in legislation.” He offers as evidence against this the observation that slavery did not arise through legislation. I confess that the appeal to the example of slavery does not come as a surprise to me. I purposely employed the word “almost” in my assertion advisedly. A spontaneous legal order does not guarantee and will not supply a completely just legal order. As I mentioned in my original posting, at any point in time, it will be rife with injustice. Advocating for a spontaneous legal order entails abandoning utopian aspirations. In the real world, the relevant question is always the comparative one: Compared to what? But in this context, Mr. Sandefur’s example not only is not a refutation of my assertion, it is strongly supportive of it.

Mr. Sandefur states, “To take only the limited example of race relations in the United States, one finds that rules designed to exploit or oppress minorities usually did not begin in legislation but were codified in legislation only after a long and dreadful life as social prejudices.” Indeed. When was racial oppression codified in legislation? When it could no longer be sustained without it.

During the “long and dreadful life as social prejudices” when slavery was widely considered an acceptable practice, market and spontaneous ordered legal forces supported it. During that time, I didn’t notice much legislation enacted to end the exploitation. However, in the 19th century, when cracks began to appear in the legal structure upholding the institution of slavery due to the increasing erosion in people’s belief in its legitimacy, national legislation such as the Fugitive Slave Act and nationally binding rulings such as that of the Dred Scott case began to appear to shore it up. The abolitionists that Mr. Sandefur refers to significantly influenced public opinion regarding the legitimacy of slavery, but they were notable for their lack of success on the legislative front where they were opposed by entrenched political interests that benefited from slavery’s existence.

Following the Civil War, the legislation that flowed from the “constructed” legal orders of the states provided an additional century of racial oppression that was completely unattainable through market and common law forces. The surest sign of what a spontaneous order produces can be read off the face of contemporary legislation. Legislation, the embodiment of collective choice, is always enacted to achieve what cannot otherwise be achieved through individual action — in modern terminology, to correct “market failures.” Jim Crow legislation was enacted for precisely this reason — to address the difficulty in attaining effective racial oppression in the realm of individual choice.

A customary/common law spontaneous legal order can contain unjust or oppressive elements as long as the vast majority of the relevant community find them morally acceptable. Of course, under that assumption, there is utterly no prospect that a representative government will improve the situation through legislation. But as the consensus in support of the oppressive practices decays, a spontaneous legal order allows motivated individuals to find ways around it, and facilitates the gradual liberalization of the order. It is here that legislation usually makes its appearance as the politically powerful beneficiaries of the oppressive practice employ the mechanism of collective choice to oppose the liberalizing aspects of the spontaneous order. I am glad that the political struggle against official racial oppression was finally won by the “rational constructivists” that Mr. Sandefur praises so highly, but that victory can hardly justify support for the constructed legal order that gave us the decades of oppression that preceded it.

Mr. Sandefur writes,

Prof. Hasnas is simply saying that because bad rules will lead to bad results, people will try to implement new rules. Obviously that’s so, but saying “well, it’ll all work out in the end” does nothing to help those who are suffering here and now. Injustice will end eventually, but only if we take the steps necessary to end it.

This suggests that he has missed my point entirely. I think it is fairly clear that I have not advocated doing nothing to help those who are suffering or to end injustice. Indeed, we should all work as hard as we can to end injustice. But the question under consideration in the present context is whether a spontaneous or constructed legal order is normatively preferable. Will advocacy against injustice be more likely to be successful in system of rules that evolve without any identifiable human agency having the power to impose a decision on the entire order or in one in which there is such an agency? I have argued for the former. My position is that advocacy against injustice is more likely to be successful in a spontaneous legal order than in a constructed one.

I understand the allure of the latter — the temptation to swoop in with the power of legislation to right wrongs and eliminate injustice. I also find the image of Don Quixote inspiring. But I believe both to be fantasies. The mechanism of collective choice that allows one to dream of achieving justice now and once and for all equally lends itself to the achievement of exploitative ends, and the incentives in constructed orders favor the latter.

Advocating against injustice in an open-ended spontaneous legal order can be unsatisfying in that it often requires a protracted effort and produces only partial success. One has to accept that progress will be gradual and incremental, and that the ideal of justice cannot be rapidly achieved, but can be heartened that the progress is likely to be sustained. One advocates for a continuously evolving customary/common law legal order not because it is ideal, but because human experience teaches that we need a prophylactic against the temptation to use the power of collective choice to achieve our ideals.

The power to legislate gave us Jim Crow. Thurgood Marshall and Charles Hamilton Houston fought a decades long campaign to gradually undermine and destroy the injustice of this legislation through private lawsuit. I would not characterize their efforts as “doing nothing.” But I would suggest that their efforts might not have been necessary in a purely customary/common law legal order in which the power to pass such legislation in the first place does not exist.



[1] See John Hasnas, Hayek, the Common Law, and Fluid Drive, New York University Journal of Law & Liberty 1:79 (2005).

[2] Mr. Sandefur correctly points out with his Wal-Mart example that constructed orders can permit decentralization and the formation of internal markets to some extent, and so can consist of a mixture of spontaneous orders within a larger constructed order. This is, in fact, the state of most real-world constructed orders since true totalitarian control is usually impossible. Nevertheless, the larger order remains an constructed one, and is easily identifiable as such.

[3] Once again, Mr. Sandefur can rightly point out that there are no pure laissez-faire markets, and that in the real world large areas of the market are carved out by government in which winners and losers are indeed determined by collective choice. But also once again, there is no difficulty identifying the portion of the market that remains free as a spontaneous order.

[4] Although, like Wal-Mart, there are large areas that are consciously permitted to remain free of direct control.

Also from this issue

Lead Essay

  • In his lead essay, lawyer and legal theorist Timothy Sandefur proposes that Friedrich Hayek’s understanding of law and justice is flawed: Spontaneous order may be a descriptively accurate concept, but it has little or no effective normative content. Depending on how one chooses to focus, those who wish to reform a spontaneous order are either constructive rationalists — thus, outside the order, and presumptively bad — or they are manifestations of the spontaneous order itself, which changes over time. He suggests that the Hayekian approach to legal reform is simply “be careful,” and that this is not terribly helpful advice.

Response Essays

  • In his response essay, John Hasnas offers solutions to Sandefur’s problems. He suggests that genuine spontaneous orders can be recognized as having no final decision makers, and hence as recognizing a multitude of individual choices. Constructed orders have a final decision maker, and do not respect individual choice. The normative benefits of a spontaneous order are therefore clear: It offers a greater scope for peaceful cooperation, while tending to reduce coercion incrementally. Still, Hasnas admits, spontaneous orders will always be “riddled with injustice,” in part owing to our own limited knowledge and virtue. He suggests that one key missing insight helps rescue much of Hayekian legal thought: the notion that laws, too, respond to market forces.

  • Daniel Klein argues that much of the fuzziness in Hayek’s writing was strategic — designed to bring lapsed liberals back into the fold, or to appeal to people who would never accept an unvarnished liberalism. Still, Klein finds great value in Hayek’s work. He argues that, while out of fashion at the time, Hayek’s own willingness to be indeterminate, and to embrace indeterminateness, was both consistent with the Smithian understanding of the social order — and predictive of some of the best work being done today in economics and in other social sciences.

  • Bruce Caldwell proposes two solutions to Sandefur’s problems. The first is to acknowledge that Hayek was a rule utilitarian, albeit one who recognized that the rules we have inherited are the products of a spontaneous order. The second is to claim that Hayek wasn’t proposing any normative conclusions at all — he was simply making observations in a value-neutral way, as might befit a member of the Austrian School, which was deeply influenced by Max Weber’s ideal of a value-neutral social science. Both approaches have their strengths and weaknesses, Caldwell admits, yet in the end he cannot accept Sandefur’s claim that there is no meaningful distinction between spontaneous and designed orders. Although the difference can be difficult to put into words, we know them when we see them.